Project Summary/Abstract Language use in ASD is extremely heterogeneous, ranging from age-appropriate to nonverbal. The PI has tracked language development in three cohorts of children with ASD (total N = 42) for the past 8-12 years, from initial diagnosis at approximately 2.5 years to follow-up at 5-8 years, with typically developing (TD) controls (total N = 45) language-matched at study onset (the Longitudinal Study of Early Language; LSEL). We demonstrated the utility of fine-grained measures of production and comprehension in characterizing individual variation in both groups. These children are now in middle childhood to adolescence, allowing us to study language use appropriate to this age, its relationship to `individual' early child and parent language measures from the LSEL and `interactive' measures drawn from parent-child conversational dynamics. This project investigates three questions about the language of school age children with ASD or TD: How does variability in children's grammatical, semantic, and pragmatic language usage during school age demonstrate the inter-dependence vs. distinctiveness of these language areas? We hypothesize that composite measures of grammar and semantics collected in the current project will manifest distinctiveness, with little overlap in variability. Composite measures of semantics and pragmatics will manifest stronger inter-dependence. (2) Which individual early child and parent measures predict children's school age language use? We hypothesize that early child measures of grammar, semantics, and social cognition will predict school age measures in the same domains (e.g., early grammatical processing will predict grammatical usage at school age). We also hypothesize that parents who used more diverse words and grammatical constructions during earlier time periods will have children with more advanced categorization and grammatical usage at school age, and that parents who used more decontextualized language and provided more narrative scaffolding during earlier time periods will have children who use more sophisticated narratives at school age. We expect that early parent measures will exert relatively more influence on later language in the TD vs. ASD group. (3) Which early and concurrent measures of interactive conversational behavior will vary by group and predict children's school age language use? We examine the relationships between interactive dynamics during play sessions gathered when the children were preschoolers, and their school age language. We hypothesize that turn-taking and backchanneling will be more prevalent in TD dyadic interactions than in interactions with children with ASD while alignment measures will vary less by group. Moreover, conversational dynamics measures will predict aspects of children's subsequent speech and comprehension levels, as well as their narratives during school age. Finally, conversational dynamics measures will mediate the predictive power of individual measures, possibly in interaction with diagnosis.